How Do Credit Rating Agencies Affect Relationship Lending?

Abstract

Small businesses have long relied upon banks to produce sufficient information to be able to profitably lend to them. In recent years information production has also taken place among information service bureaus, such as Dun and Bradstreet (D&B). There has been some evidence to suggest that these information brokers have allowed small firms to break away from their lending relationships.In this paper, we present a model of how small business loans are analyzed, focusing on the information that is necessary to produce. Based on the insight from that model, we build a multinomial logistical regression model to estimate joint probabilities that a firm with at least one line of credit will fall into certain categories of D&B credit rating and secured/unsecured line of credit. From this multinomial logistic regression model, we derived conditional logistic regression, where the probability the line of credit is secured is conditioned upon the credit rating category.We find that the length of a relationship with the line of credit lender is only a significant predictor of collateral for firms with lower credit ratings, consistent with the theory that relationships are needed to overcome informational opacity. In fact, it appears from the set of logistic regressions run, that the significance of the length of the lending relationship to the granting of unsecured lines of credit reported in earlier research is driven solely by firms with lower credit ratings. The length of the relationship was not significant in assessing the probability of secured lines of credit for those with the highest credit rating.